Exact Solution for Stimulated Loss

Why Constant Stimulation of Loss Has an Exact Solution

When a drug stimulates response loss, the response tends to fall below its untreated baseline. The exact trajectory depends not only on the lower equilibrium but also on how quickly the system approaches it. Under constant stimulation, both the production rate and the first-order loss-rate coefficient are constant. The loss flux itself is not constant: it remains proportional to the response, which changes over time. This structure gives a first-order linear differential equation with a closed-form solution.

The result is useful for calculating response at a specified time, determining how long it takes to approach a target, and checking what a constant-exposure approximation implies. It is not a claim that drug exposure or stimulation is constant in every dosing situation.

Definition: Let $R(t)$ be the response at time $t$, $k_{\mathrm{in}}>0$ the constant production rate, and $k_{\mathrm{out}}>0$ the untreated first-order loss-rate coefficient. Let $S\geq 0$ be a constant, dimensionless stimulation of loss. The stimulated-loss model is $$ \frac{dR(t)}{dt}=k_{\mathrm{in}}-k_{\mathrm{out}}(1+S)R(t),\qquad R(0)=R_i. $$ Here $R_i$ is the initial response. The untreated equilibrium is $R_0=k_{\mathrm{in}}/k_{\mathrm{out}}$. The coefficient $k_{\mathrm{out}}(1+S)$ has units of inverse time; the production rate has units of response per time.

Equilibrium and Exact Trajectory

Write $\lambda=k_{\mathrm{out}}(1+S)$, the stimulated first-order loss-rate coefficient. Because $k_{\mathrm{out}}>0$ and $S\geq0$, $\lambda>0$. At equilibrium the response is not changing, so setting $dR/dt=0$ gives

0=k_{\mathrm{in}}-\lambda R_{\mathrm{eq}} \quad\Longrightarrow\quad R_{\mathrm{eq}}=\frac{k_{\mathrm{in}}}{k_{\mathrm{out}}(1+S)} =\frac{R_0}{1+S}.

Thus, for $S>0$, the equilibrium is below the untreated baseline. The factor $1/(1+S)$ describes the equilibrium reduction for this linear stimulation form. The coefficient $\lambda$, meanwhile, determines the rate of approach to that equilibrium.

To derive the trajectory, define the deviation from equilibrium as $D(t)=R(t)-R_{\mathrm{eq}}$. Since $k_{\mathrm{in}}=\lambda R_{\mathrm{eq}}$, the differential equation becomes

\frac{dD(t)}{dt} =\frac{dR(t)}{dt} =k_{\mathrm{in}}-\lambda R(t) =\lambda R_{\mathrm{eq}}-\lambda R(t) =-\lambda D(t).

If $D(0)=0$, then the response starts at equilibrium. The constant function $D(t)=0$ solves this equation, so $R(t)=R_{\mathrm{eq}}$ for all $t$. If $D(0)\ne0$, separation of variables is valid, and integration gives

\int_{D(0)}^{D(t)}\frac{1}{D}\,dD =-\lambda\int_0^t du \quad\Longrightarrow\quad \ln\!\left(\frac{D(t)}{D(0)}\right)=-\lambda t.

Exponentiating and substituting $D(0)=R_i-R_{\mathrm{eq}}$ yields the solution. It also holds in the separate case $D(0)=0$, because then the exponential term is zero.

Theorem: For $k_{\mathrm{in}}>0$, $k_{\mathrm{out}}>0$, constant $S\geq0$, and any initial response $R_i$, the solution is $$ R(t)=R_{\mathrm{eq}}+(R_i-R_{\mathrm{eq}})e^{-k_{\mathrm{out}}(1+S)t}, \qquad R_{\mathrm{eq}}=\frac{R_0}{1+S}, \qquad t\geq0. $$ The exponential is dimensionless because its exponent is a rate multiplied by time. The deviation from equilibrium has the same sign as its initial value and approaches zero as time increases.

The characteristic time constant is $1/\lambda$, and the time for the deviation from equilibrium to be halved is $\ln(2)/\lambda$. These are properties of the response trajectory under the stated constant-coefficient model; they are not, by themselves, estimates of a drug’s plasma half-life.

Worked Calculations

Worked Example: Response After One Hour

Suppose a hypothetical system has $k_{\mathrm{in}}=12$ response units/hour, $k_{\mathrm{out}}=0.60$ per hour, and constant stimulation $S=1.50$. Its initial response is $R_i=20$ response units. Calculate the response after one hour.

The untreated baseline and stimulated loss-rate coefficient are $$ R_0=\frac{12}{0.60}=20\ \text{response units}, \qquad \lambda=0.60(1+1.50)=1.50\ \text{per hour}. $$ The stimulated equilibrium is $R_{\mathrm{eq}}=20/(1+1.50)=8.00$ response units. Substituting $t=1$ hour into the solution, $$ R(1)=8+(20-8)e^{-(1.50)(1)} =8+12e^{-1.50} =8+12(0.2231301601\ldots) \approx10.68\ \text{response units}. $$ The response has moved toward the lower equilibrium but has not reached it after one hour.

Worked Example: Response at a Later Time

For a different hypothetical system, let $k_{\mathrm{in}}=8.00$ response units/hour, $k_{\mathrm{out}}=0.40$ per hour, and $S=0.50$. If $R_i=30.0$ response units, find the response after three hours.

The equilibrium and coefficient are $$ R_{\mathrm{eq}}=\frac{8.00}{0.40(1.50)} =13.3333\ \text{response units}, \qquad \lambda=0.40(1.50)=0.60\ \text{per hour}. $$ Therefore, $$ R(3)=13.3333+(30.0-13.3333)e^{-(0.60)(3)} \approx13.3333+16.6667(0.1652988882) \approx16.09\ \text{response units}. $$ The response remains above its equilibrium because it began above equilibrium, and the positive deviation has shrunk.

Worked Example: Time to Reach a Target Response

In a hypothetical model, $k_{\mathrm{in}}=9.00$ response units/hour, $k_{\mathrm{out}}=0.30$ per hour, and $S=2.00$. The initial response is $4.00$ response units. How long does it take to reach $8.00$ response units?

First calculate $$ R_{\mathrm{eq}}=\frac{9.00}{0.30(3.00)}=10.0\ \text{response units}, \qquad \lambda=0.30(3.00)=0.90\ \text{per hour}. $$ Set the trajectory equal to the target and solve: $$ 8=10+(4-10)e^{-0.90t} \quad\Longrightarrow\quad 2=6e^{-0.90t} \quad\Longrightarrow\quad e^{-0.90t}=\frac{1}{3}. $$ Thus $$ t=\frac{\ln(3)}{0.90}=1.22068\ldots\ \text{hours} \approx1.221\ \text{hours}. $$ The target lies between the initial response and equilibrium, so the calculated positive time is consistent with the trajectory.

Assumptions, Diagnostics, and Limits

The derivation requires a constant production rate and a constant stimulated loss-rate coefficient over the interval considered. It also assumes loss is first order in response, stimulation multiplies $k_{\mathrm{out}}$ by $1+S$, and the model parameters do not change during that interval. The initial response can be any value for which the model is meaningful; it need not equal the untreated baseline. If the response is a nonnegative quantity, the usual application also requires a nonnegative initial value.

Several features provide direct checks on calculations and fitted trajectories:

Direction: If $R_i>R_{\mathrm{eq}}$, the response decreases toward equilibrium; if $R_i<R_{\mathrm{eq}}$, it increases toward equilibrium. It does not cross the equilibrium under this constant-coefficient solution.
Initial slope: Differentiating the solution at $t=0$ gives $dR/dt=-\lambda(R_i-R_{\mathrm{eq}})$. Its sign must point toward equilibrium.
Late-time behavior: As $t$ increases, $e^{-\lambda t}$ approaches zero and $R(t)$ approaches $R_{\mathrm{eq}}$. Persistent drift away from that value is inconsistent with the assumed constant model.
Scale of time: Larger $S$ increases $\lambda$, shortening the time constant $1/\lambda$. It also lowers $R_{\mathrm{eq}}$. These are distinct effects of the same stimulation parameter.

In observed data, measurement noise can obscure these patterns. Systematic curvature that cannot be captured by a single exponential, a changing apparent rate of approach, or a response that moves away from the predicted equilibrium may indicate changing exposure or stimulation, an incorrect equilibrium, or a model structure that is too simple. Such deviations are diagnostic clues, not proof of a particular alternative mechanism.

For a constant coefficient, the exact formula is preferable to numerical integration because it is direct and avoids numerical approximation error. Numerical integration can be useful when the input or coefficients vary with time, but the constant-stimulation formula should not be applied unchanged to such a situation. Compared with stimulated production, which changes the input term, stimulated loss changes the coefficient multiplying the response. That distinction is why the equilibrium and trajectory must be derived from the loss equation rather than inferred from the name of the drug effect alone.

Common Pitfall

Key takeaway: Constant stimulation makes the first-order loss-rate coefficient constant; it does not make the loss flux constant. The flux is $k_{\mathrm{out}}(1+S)R(t)$, so it changes whenever $R(t)$ changes. The exact solution accounts for that response dependence.

Another error is to use the untreated equilibrium $R_0$ as the asymptote during stimulation. Setting the derivative to zero under stimulation gives $R_{\mathrm{eq}}=R_0/(1+S)$, not $R_0$. Finally, the logarithmic separation used in the derivation is valid only when the initial deviation is nonzero. When $R_i=R_{\mathrm{eq}}$, the deviation stays zero and the solution is simply constant at equilibrium.

Check Your Understanding

Answer each question, then compare with the solutions below.

If $R_0=24$ response units and $S=1$, what is the stimulated equilibrium?
For constant stimulation, what parameter controls the exponential rate of approach, and what are its units?
A response starts below its stimulated equilibrium. Does the model predict that it crosses the equilibrium? Explain.
What happens when the initial response equals the stimulated equilibrium, and why must this case be handled separately in the separation-of-variables derivation?

Solutions. (1) $R_{\mathrm{eq}}=R_0/(1+S)=24/2=12$ response units. (2) The rate is $\lambda=k_{\mathrm{out}}(1+S)$, with units of inverse time; it appears in $e^{-\lambda t}$. (3) No. The solution is $R(t)=R_{\mathrm{eq}}+(R_i-R_{\mathrm{eq}})e^{-\lambda t}$, and the exponential is positive, so the deviation retains its initial negative sign while shrinking toward zero. (4) The response remains at equilibrium, $R(t)=R_{\mathrm{eq}}$. In that case $D(0)=0$, so dividing by $D$ and taking $\ln(D(t)/D(0))$ is undefined; the constant solution establishes this case directly.